Part I - New Developments Labo: 
iorit^ 
Image Quality value 
phys 
During the period since the 1960 Congress, interest has continued and many lab- t00 * 
oratories have been active in the area of image-evaluation, with the prime object- ^ as ^ 
ive of expressing micro-image quality in more fundamental ways than the long some 
established re solving-power. Fourier theory has provided the principal tools in this ever 
attack on a long-standing problem, transfer functions being used to describe the on 
image characteristics of lenses, emulsion, image motions, etc. , while the non- imag 
linearities of the emulsion are expressed in terms of the H & D curve. The addition or v£ 
of granularity data to characterize the "noise" introduced by the developed emulsion accu 
completes what appears to be a plausible mathematical model of the physics of 1S jn 
photographic imagery. In fact, what may be termed the "Fourier model" has been cond 
very useful to the research worker and the system designer, and has greatly improv 
ed our insight into the complexities of the photo-optical system. Hempenius , for othei 
example, gives an excellent account of this approach to system design.* From the enou 
point of view of the practical user, however, and those seeking a simple test, whic 
better than re solving-power, the activity has been singularly unprofitable. The 
reasons for this lie in the difficulty of making the necessary measurements and ge 
the problems of interpreting them. spat: 
caus 
of pt 
The measurement of transfer functions, whether of lenses or emulsions, is an al- esse 
together more formidable undertaking than doing a resolving-power test, and is form 
only possible for well-equipped laboratories using highly precise and necessarily sumr 
expensive equipment. (The questionnaire replies provided ample confirmation of 
this) While such measurements are within the scope of the larger organizations, 
such as lens and emulsion manufacturers, universities, and national testing lab 
oratories, etc. , it will never be possible for the general user to measure a transfer 
function in the way he can now make a re solving-power test. The fuller information 
provided by the transfer function has to be paid for in the sophistication of its t 0 
measurement. of 
the 1 
A proper appreciation of the value and limitations of transfer functions has been nega 
delayed by misunderstandings as to their nature, particularly in reference to 
emulsions. By definition the transfer function expresses image modulation as a Thou 
function of spatial frequency for sinusoidal targets, but this seems to have been naire 
widely misunderstood to express image contrast as a function of reciprocal image thou 
size for any shape of target. The contrast of images in general can be derived from by s 
the transfer function, but only by using the latter for its legitimate purpose, i.e. , to came 
operate on a spatial frequency spectrum. Since the frequency spectra of targets in move 
general are not known (though they can be derived) the transfer function by itself is ment 
not a guide to image quality. Methods are used whereby an approximate indication Q f th 
of re solving-power can be derived from the transfer function, but for the general char 
user this is hardly an improvement on direct measurement. the i 
qua! 
* Hempenius, S. A. Applied Optics, Vol. 3, No. 1, Jan 1964. comr 
func 
Prof« 
that